|From:||"John Thalakkottor,Joseph" <tjjoseph@...1447...>|
|Date:||Wed, 29 Nov 2017 08:39:27 +0000|
I am interested in evaluating local fluid stress using Lammps and going through the mailing list archives found that using the 'compute stress/atom' in combination with 'fix ave/spatial' might not yield the best results and its more appropriate to use 'fix atc hardy' to evaluate local stress components.
In order to understand the fix atc package and validate its output, I performed a simple Couette flow simulation and try to compute viscosity from shear stress and shear rate. The velocity profiles from both 'fix atc' and 'stress/atom+fix ave/spatial' match, while the shear stress don't, as expected. On computing the viscosity the results using 'stress/atom+fix ave/spatial' give viscosity value of 2.3, which is closer to the literature (2.1, Thompson and Troian, Nature 1997) as compared to 'fix ATC' which gives a value of 0.38. Also to note is that while evaluating shear stress from 'stress/atom+fix ave/spatial' I am dividing the results by the 'volume of the bin/number of particles in the bin'.
I am bit confused now and am trying to figure out what I am doing wrong and which is the more appropriate way to compute fluid stress.
Please find attached the log file, data file and the velocity and stress plots.
Any help or suggestion is much appreciated.